Johnny Manziel is also really good at football

Two days ago I did some analysis on Mike Evans' performance against Alabama. It was good (Evans, not the analysis . . . though the analysis was pretty good, too). I promised a follow up on Manziel, and I deliver:

Manziel threw for 464 yards on 39 attempts and rushed for another 98 on 14 carries (a sack cost him 5 yards, so Manziel exceeded 100 "rushing" yards). Given the opposition, this marks the best offensive performance in the history of college football . . . if it weren't for two interceptions.

Yards
Manziel totaled 562 yards on Saturday; that's the 19th most yards by a player in a game since 2005. In those 8 1/4 seasons, the 550 mark has been eclipsed 25 times. Manziel has done it three times, Colt Brennan did it twice, and 20 others did it once.

But we should also account for the quality of the opposition. Geno Smith, for example, totaled 687 yards against Baylor, but Baylor allowed over 500 yards per game in 2012. To account for this, next to the yards I've listed the average yards per game allowed by each opponent. In Manziel's case, I used Alabama's 2012 average yards per game, but I substituted A&M's 628 yard game Saturday for their 418 yard game against Alabama last year; Alabama's average yards allowed increased from 250 to 265 as a result. In the final column, I divided the two: Yards/Average yards allowed.

When we do this, Manziel's game against Alabama comes out on top. After his 2.12, Colt Brennan is a distant second with 1.75 followed by Cody Hodges in 2005 with a 1.68. In fact, Manziel's 2.12 is the biggest number not just on this list but on any list. When we look at all players in all games, the second highest yards/average yards allowed ratio is 1.82 from Chase Holbrook against Boise State in 2005 followed by Geno Smith's 1.79 against LSU in 2011. Manziel more than doubled the average yards allowed per game of his opponent; no one else has come close . . . at least recently . In short, Manziel's 562 yards was the most impressive yardage total of the last 9 years, and not by a little.

EPA

************************************************

The best stat for measuring an individual players' production (which, I admit, is different than performance, but we'll get back to that) is the EPA, or expected points added. Essentially, it looks at where a team was before a play, in terms of score and field position, and after a play and assigns a point value to the play. That point value is based on how the score and field position changed. Field position, down and distance are converted to an expected point value based on what teams do on average over the course of the possession from that situation. 

If that doesn't make any sense, remember this: the higher the EPA the more a player did to help his team score more points and make it harder for the other team to score points.

************************************************

Against Alabama, Manziel scored a 22.6. That's good. It is the 15th highest single-game EPA by a player this season (AJ McCarron scored a 23 in that same game). But it could have been much better. Sunseri grabbed the tipped pass and took it back for a touchdown. On that one play Manziel scored a -7.2. Remove that play and Manziel scores a 29.8 and the 4th highest single-game EPA of the season (2 points behind Mariota vs. Tennessee). If Manziel had managed to bring down Sunseri after the interception he finishes with a 26.4 (8th best).

We can't just remove bad plays and highlight good plays, but we also shouldn't compare apples and oranges. The three highest EPA performances this season came from Mariota (vs. Tennessee), Blake Bell (vs. Tulsa) and Keeton (vs. Air Force). Each performance is impressive in its own right, but racking up stats against Tennessee, Tulsa and Air Force, or even McCarron's 23 against A&M, is not the same as doing it against Alabama.

Fortunately, we can adjust a player's performance for the strength of the competition. Basically, we look at what other players have done against that defense and against all other defenses, throw it in a fancy algorithm, and come up with a value that reflects how hard it is to run, throw and catch against each team.

When we adjust for the competition, Manziel's 22.6 becomes a 50.5, the highest-single game EPA+ ("+"=adjusted) of the season (Mariota's 40.8 against Tennessee is tied for second). But we shouldn't put too much stock in these results. Manziel has only played one half against another FBS opponent and Alabama had only played Virginia Tech. The 50.5 is the computer's way of saying, "Holy smokes, this Manziel guy is a lot better than Logan Thomas." That's not news.

So instead, I threw this game in the 2012 season - what if 2012 Johnny Manziel had done this against 2012 Alabama? How would the computer interpret that? When we do that, and adjust for competition, Manziel against Alabama comes in 22nd since 2005. Great, but not historical. Sam Bradford had three better games, RGIII, Colt Brennan, and Vince Young had two. Even Manziel himself scored higher against Missouri last season. Geno Smith against Baylor, the leader of the pack, was 5.7 points better (Smith's raw EPA from that game was a 51.3, almost 30 points higher than Manziel's). This is not an insignificant gap, but if that tipped ball had somehow fallen through Sunseri's hands, Manziel would have had the single-most productive game for an individual player in the last nine years . . . and probably of all time. Throw out the second interception and this performance would be a full touchdown better than any we've ever seen before. Unfortunately, football doesn't work like that.

Subjectively, given the moment, I've said that I think this was the most impressive offensive performance in the last decade and maybe of all time (though I'm admittedly too young to speak of such things). The accumulated stats with the (insufficient) 4th quarter comeback are awe worthy. If the A&M defense had come up with a couple more stops, the rest of the college football world might see it the same way.